A conversation series with Franklin De Vrieze on democratic institutions, accountability and what happens after decisions are made.
How do democracies learn? How do institutions respond when laws fail, systems are tested, and new challenges emerge?
In this new DT500 MAG series, critic Arthur Sopin speaks with Franklin De Vrieze, Head of Practice Accountability at Westminster Foundation for Democracy, about what happens beyond elections and political decisions – the processes of scrutiny, implementation, institutional learning and democratic resilience that determine whether decisions actually work in practice.
From post-legislative scrutiny and parliamentary oversight to AI, cybersecurity, war and authoritarian power, the series asks a simple question:
01 – CAN DEMOCRACY LEARN FROM ITS MISTAKES?
What happens after a law is passed – and what happens when democratic institutions fail to learn from what comes next?
CAN DEMOCRACY LEARN FROM ITS MISTAKES?
What happens after a law is passed – and what happens when democratic institutions fail to learn from what comes next?
Democracy is often reduced to a moment.
Every four or five years, citizens are asked to cast a ballot. Governments are formed. Members of parliament are elected. Leaders make promises.
Then the attention moves elsewhere.
But what happens after the vote?
What happens after parliament has debated a bill, passed it into law and moved on to the next political battle?
For Franklin De Vrieze, that is where some of the most important questions about democracy begin.
“Democracy is much more than elections,” he argues. It is about the political process that follows — and the relationship between citizens and the institutions acting on their behalf.
At the heart of that relationship is what he describes as the social contract.
Citizens elect representatives and expect them to act in the public interest. In return, political institutions are expected to operate responsibly, accountably and without corruption.
The contract, however, is not self-enforcing.
It has to be maintained.
And that means asking a question that democratic systems are often surprisingly bad at asking:
Did the thing we decided to do actually work?
THE SECOND LIFE OF A LAW
There is a widespread assumption that once parliament passes legislation, the difficult part is over.
A bill is debated. Parliament votes. It receives formal approval. It enters into force.
But legislation does not implement itself.
There may be insufficient funding. Officials may not have been trained. Regulations necessary to put the law into practice may be delayed. Responsibility for implementation may be unclear.
And even when all of those things happen, legislation may produce consequences its authors never anticipated.
De Vrieze has spent much of his career examining this space between legislation and reality.
The process is known as post-legislative scrutiny: assessing what happens after legislation has been enacted and asking whether it is working as intended. Westminster Foundation for Democracy has developed extensive work on the subject, including a parliamentary guide authored by De Vrieze and a later updated manual on parliamentary innovation through post-legislative scrutiny.
The idea sounds technical.
Its implications are not.
Because without systematic scrutiny, governments can respond to almost every social problem with the same instinct:
pass another law.
But if the existing law has never been properly evaluated, how do you know whether a new law is necessary – or whether it will actually produce better results?
And how do you know whether the old one failed because of its design — or because nobody properly implemented it?
DEMOCRACY NEEDS A MEMORY
De Vrieze describes legislation not as a single act but as a cycle.
A problem is identified.
A policy is developed.
Legislation is drafted.
Parliament debates and adopts it.
The law is implemented.
Courts and public authorities apply it.
Then comes the stage that is too easily forgotten:
What did we learn?
For De Vrieze, institutional learning should not begin only after a disaster.
Too often, politicians become interested in “lessons learned” after a scandal, accident or investigative journalist exposes a failure.
By then, the damage has already happened.
A mature democratic institution should not need catastrophe to discover its own blind spots.
It should be capable of looking for them.
That means examining unintended consequences, identifying gaps and listening to people who may never have been powerful enough to shape the original legislation.
It also means accepting something that political institutions can find uncomfortable:
the possibility that they were wrong.
THE PEOPLE WHO WERE NEVER IN THE ROOM
One of the clearest examples of post-legislative scrutiny comes from the UK’s Equality Act 2010.
In 2015, the House of Lords established the Equality Act 2010 and Disability Committee to conduct post-legislative scrutiny of the Act’s disability provisions. Rather than attempting to scrutinise the entire Act, the committee focused on its impact on disabled people and examined whether the legislation adequately supported the fight against disability discrimination.
That distinction matters.
The committee examined issues including the public sector equality duty, access to buildings and transport, access to justice and the role of the Equality and Human Rights Commission.
Its report identified areas where disabled people were being poorly served and made recommendations for change.
The transport provisions offered a particularly revealing example.
Section 165 of the Equality Act was intended to place duties on drivers of designated taxis and private-hire vehicles to carry wheelchair users, provide reasonable assistance and, in specified circumstances, carry the wheelchair without charging the passenger more.
Yet important provisions had not initially been brought fully into force.
The problem, then, was not simply whether Parliament had legislated for equality.
The question was whether the legislation had actually translated into protection in people’s everyday lives.
The committee’s scrutiny exposed the gap between legislation on the statute book and its implementation in practice. It demonstrated why post-legislative scrutiny is different from simply passing legislation: Parliament had to return to an existing law, examine what had happened after its enactment and ask what still needed to change.
The distinction is at the heart of De Vrieze’s argument.
A law sitting on paper is not the same thing as a law working in society.
NOT REPORTS FOR THE ARCHIVES
This is also why De Vrieze rejects the idea that institutional learning is simply another report.
A parliamentary inquiry can produce hundreds of pages.
But a report that disappears into an archive has not necessarily changed anything.
The important question is what follows.
Does the evidence influence policy?
Does government respond?
Are gaps addressed?
Are previously unheard groups brought into the political conversation?
And does parliament return to the issue later to see whether anything actually changed?
For De Vrieze, scrutiny should therefore be understood as a mechanism for correction, not merely documentation.
That distinction matters because parliament has something government ministries do not possess in quite the same way: a direct relationship with competing political voices and with citizens.
Parliamentary committees can hear from experts, civil society organisations, affected communities and members of the public.
The UK’s petition system offers another route into this relationship.
British citizens and UK residents can create petitions. Petitions receiving 10,000 signatures receive a government response, while those reaching 100,000 signatures may be debated in Parliament.
It is a reminder that democratic participation does not necessarily begin and end at the ballot box.
THEN CAME THE WAR
If the UK examples demonstrate the mechanics of institutional learning, Ukraine demonstrates why those mechanisms can matter under extraordinary pressure.
De Vrieze has worked with Ukrainian parliamentary institutions during wartime.
What surprised him most was not simply that the Ukrainian parliament continued to exist.
It was that it continued to innovate.
Parliamentary committees have continued to adapt their work through digital and hybrid forms of engagement, while parliamentary institutions and civil society have continued to develop mechanisms for public participation during wartime.
Technology has become part of the infrastructure through which democratic institutions continue to function.
But De Vrieze’s most powerful memory is not technological.
It is physical.
During a conference in a Ukrainian border town, an air-raid alarm sounded.
The participants moved into the basement.
The conference continued underground.
Later, there was another alert.
This time, an explosion occurred close enough for those in the basement to feel the vibrations.
Then the meeting continued.
What stayed with De Vrieze was the determination of his Ukrainian colleagues.
They were not simply trying to preserve a parliamentary procedure.
They were trying to preserve the idea that political institutions should continue to function even when the conditions around them are collapsing.
THE AIR-RAID SHELTER
There is something almost symbolic about the scene.
Democracy reduced to its physical essentials.
A group of people underground, continuing a political discussion while war continues above them.
It challenges an assumption that democratic institutions require perfect conditions in order to function.
Ukraine suggests otherwise.
Post-legislative scrutiny, De Vrieze argues, should not be treated as a luxury available only to wealthy, stable democracies with highly resourced parliaments.
It is part of parliament’s fundamental responsibility:
oversight and accountability.
That is why institutional learning matters.
Not because democratic institutions are incapable of making mistakes.
But because they must have mechanisms through which mistakes can be discovered and corrected.
THE AI PROBLEM
And now the problem is becoming more complicated.
Artificial intelligence is changing the speed at which institutions can process information.
Parliaments are already exploring AI for tasks including research, analysis and administrative work. Westminster Foundation for Democracy’s Guidelines for AI in Parliaments, for which De Vrieze was an editor, take a broad approach covering ethics, privacy, security, oversight, system design and education.
But technology creates a new asymmetry.
What happens if governments acquire AI capabilities faster than parliaments?
What happens if ministries and public agencies begin accumulating enormous quantities of data while legislators responsible for scrutinising them lack comparable analytical capacity?
The “pacing problem” describes a familiar challenge in technology governance: technological change can move faster than the ability of political and regulatory institutions to understand, govern and respond to it.
For democratic institutions, the concern is particularly acute when the executive can adopt and deploy new technological capabilities faster than the legislature can develop the capacity to understand and scrutinise them.
The executive moves faster than the legislature.
A government that can process vast amounts of information through increasingly sophisticated systems may become significantly more powerful relative to a parliament that cannot keep pace.
That is not simply a question of whether MPs know how to use AI.
It is a question of institutional power.
If parliament cannot understand the information available to government, how can it effectively scrutinise government?
THE PARLIAMENT THAT CANNOT SEE
This is where the future of democratic oversight becomes inseparable from technology.
De Vrieze argues that AI should not be treated simply as an IT project delegated to a parliamentary technology department.
It is an institutional transformation.
Parliaments need policies governing how data is created, accessed and analysed.
They need meaningful human oversight.
They need to decide which AI applications should be prioritised.
And because parliament is inherently political, they need broad agreement between political groups and administrative staff.
Westminster Foundation for Democracy’s AI guidelines similarly emphasise governance, oversight, data policies, privacy, security, system design and institutional capacity.
There is another problem, too:
cybersecurity.
A parliament is not simply a building.
It is a distributed network of elected representatives, constituency offices, staff, personal devices and digital systems.
That decentralisation can create vulnerabilities that a more centralised government department may not face in the same way.
Cybersecurity therefore becomes part of democratic resilience.
An attack on a parliamentary institution may be a technical incident, but its consequences can extend into the democratic process itself.
WHO TELLS PARLIAMENT IT IS WRONG?
Democratic systems contain institutions whose purpose is, in effect, to tell other institutions that they are wrong.
The National Audit Office is one example.
Its work examines public spending and whether government is achieving value for money. It reports to Parliament and provides independent analysis that can be used by MPs and committees in scrutinising departments and public services.
The principle is simple:
even democratically elected institutions can make mistakes.
That is why democracy needs internal mechanisms of correction.
Auditors.
Courts.
Parliamentary committees.
Journalists.
Citizens.
Civil society.
And sometimes the people directly affected by a policy.
The strength of a democratic system is therefore not that everyone inside it agrees.
It is that disagreement can be institutionalised without destroying the institution itself.
DEMOCRACY’S UNCOMFORTABLE ADVANTAGE
There is, however, another side to De Vrieze’s work that is less comfortable.
His work on authoritarianism examines not simply individual regimes, but the wider ecosystem through which authoritarian power can be maintained, adapted and reinforced.
The Authoritarian Ecosystem, a 2026 collection published jointly by the Westminster Foundation for Democracy and the Political Studies Association, examines interconnected institutions, actors and practices associated with the maintenance of authoritarian power. De Vrieze is among the contributors to this wider body of work.
That perspective adds another dimension to the question of institutional learning.
Authoritarian power can also adapt. Regimes can learn from one another, develop new forms of cooperation and use institutions that appear familiar — legislatures, courts, administrative systems and international networks — in ways that help consolidate power.
Dissidents can be pursued across borders.
Critical journalists can become targets.
Anti-corruption institutions can themselves be captured and weaponised.
The point is not simply that authoritarian power suppresses opposition. It is that it can operate through institutions, networks and practices that evolve over time.
The lesson is uncomfortable:
Institutions can learn in the service of democracy – but they can also learn in the service of power.
That makes democratic institutional learning something more than an administrative exercise.
It becomes part of the struggle over who controls the future.
THE QUESTION WE SHOULD BE ASKING
Democracy is often judged by its ability to make decisions.
Elections.
Votes.
Laws.
Budgets.
Governments.
But perhaps that is only the first test.
The harder test comes afterwards.
Can a parliament look at a law it helped create and ask whether it worked?
Can it listen to people who were not heard the first time?
Can it admit that implementation failed?
Can it correct itself before a scandal forces it to?
Can it keep pace with a government whose technological capabilities are accelerating?
Can it protect itself from cyberattacks?
Can it remain accountable when the information required for accountability is increasingly controlled by systems it does not fully understand?
And can it learn without waiting for disaster?
For De Vrieze, these are not peripheral questions.
They go to the heart of what democratic institutions are supposed to do.
A democracy that never changes its mind may look decisive.
But a democracy that cannot learn may simply be repeating its mistakes.
The real measure of an institution may therefore not be whether it gets everything right.
It may be whether it remembers what it got wrong.
